3.3.1 - Raine’s PET study

The question and comparison groups

Raine and colleagues asked whether people pleading not guilty by reason of insanity (NGRI) showed different cortical and subcortical brain functioning from comparison participants.

Earlier work linked impaired frontal functioning with difficulty regulating behaviour. The researchers expected differences in the prefrontal cortex and in other areas implicated in emotion and communication between brain regions. The question concerned a network, rather than a search for one “murder centre”.

The NGRI group contained 41 people: 39 males and 2 females, charged with murder or manslaughter and referred for brain examination in the context of their legal cases. Their mean age was 34.3 years. They were compared with 41 controls matched for sex and approximately for age; control mean age was 31.7 years. Six NGRI participants with schizophrenia were matched with six controls with schizophrenia. Thus “controls” did not mean that every control was free from psychiatric diagnosis.

This was a quasi-experimental comparison of pre-existing groups. Legal/clinical group membership was not manipulated or randomly allocated. The measured outcomes were regional glucose-metabolism signals. Matching made age, sex and schizophrenia less plausible simple explanations of a group difference, but could not make the groups identical in every relevant experience.

The sample must be described accurately. It did not represent all people who commit homicide, everyone with a mental-health diagnosis or people who are generally aggressive. Pleading NGRI is not the same as demonstrating that a court accepted that plea. These distinctions matter when deciding to whom the findings apply.

Measuring metabolism during a common task

The researchers gave both groups a common activity so that the PET signals could be compared under similar task demands.

Participants practised a continuous performance task (CPT) before tracer injection. The task involved detecting targets and maintaining attention. They then received a radiolabelled glucose tracer and completed the task for 32 minutes, followed by PET scanning. The scan recorded tracer uptake associated with glucose metabolism during the uptake period. Higher or lower uptake was interpreted as an indirect indicator of regional functional activity; it was not a direct measurement of aggressive thoughts.

Both groups followed the same task and imaging procedure. Participants had been medication-free for two weeks beforehand, with urine screening used as a check. The researchers compared cortical and subcortical regions and left–right activity using the paper’s cortical-peel and box analyses. These approaches organised scan data into regions that could be compared quantitatively.

The CPT engaged attention and frontal functioning, so the study tested group differences under a particular cognitive demand. CPT performance did not differ significantly between groups. This makes a simple explanation based on one group failing to perform the task less persuasive. It does not show that their everyday thinking, emotional regulation or aggressive behaviour was the same.

Standardising the task makes comparisons more interpretable, but it creates a validity trade-off. Detecting targets in controlled conditions differs from responding to a real interpersonal conflict. A metabolic difference during the CPT therefore needs further evidence before it is used to explain an act of violence.

The regional pattern of findings

The findings involved several brain regions and some left–right imbalances, rather than uniformly lower activity throughout the brain.

Compared with controls, the NGRI group showed lower glucose metabolism in prefrontal regions, the superior parietal gyrus, the left angular gyrus and the corpus callosum. The official Pearson account also reports greater occipital activity and no overall temporal-region difference. The corpus callosum connects the hemispheres; its result therefore concerns communication within a wider network as well as localised functioning.

Subcortical findings require precise wording. In the amygdala and hippocampus, the NGRI group showed lower activity on the left and higher activity on the right relative to controls. In the thalamus, right-side activity was higher, while the left-side comparison showed no difference. The original abstract summarises abnormal asymmetries in the amygdala, thalamus and medial temporal region. It would be inaccurate to reduce this to “the amygdala was less active” or “all limbic regions were overactive”.

These interpretations connect findings to existing psychological knowledge. A scan does not itself demonstrate impulsivity, lack of fear or impaired moral reasoning. Those claims need relevant behavioural measurements. Nor does a lower metabolic signal establish that the tissue was physically damaged: function and structure are different outcomes.

What the evidence can establish

The study supports a biological association with the selected NGRI group, but its strongest defensible conclusion is narrower than a causal explanation of homicide.

The authors proposed that abnormal cortical and subcortical functioning might predispose some individuals to violence. A predisposition is a vulnerability or increased tendency under some conditions, not an inevitable outcome. Biological functioning may interact with social experiences, learning and immediate circumstances. The observed network is compatible with this account; it does not establish a universal biological profile of a person who will become violent.

PET supplied quantified physiological evidence that did not depend only on participants describing themselves. The common task, matched controls and checks on medication strengthened the comparison. The reported pattern was not readily explained by the examined age, sex, schizophrenia, handedness, ethnicity or head-injury differences. These checks constrain particular alternatives, rather than demonstrating that no confounding factor could exist.

Important alternatives remain. The groups differed in legal and life histories, and the researchers did not randomly allocate biological functioning or measure it before the offences. Differences could precede violence, follow relevant experiences or reflect a third influence. The scan measured a particular task condition, while “aggression” includes many behaviours. Very few women were studied, and selection around an NGRI plea limits generalisation beyond that group. Group averages also allow overlap between individuals; they cannot classify an individual as dangerous.

Ethical evaluation concerns both the procedure and its use. Radiation and tracer injection need informed consent and justified risks. The forensic setting can create pressure, so consent must be voluntary and capacity assessed appropriately. Brain and diagnostic information requires confidentiality. Reporting that all murderers have a defective brain, or that all people with psychiatric difficulties are violent, would go far beyond the sample and risk serious stigma.

The study remains valuable because it made competing biological claims testable with living-brain evidence. Its limitations define the next questions: whether differences occur before violent behaviour, whether they predict outcomes beyond other risk factors, and whether the pattern appears in more representative samples.